AI summary
Seven AI prompts that respect what selling actually is (relationships built on signal, not template volume): account-research briefs, personalized cold outbound, discovery question stacks, objection responses, multi-thread champion mapping, close plans, and lost-deal post-mortems. Built to amplify rep judgment, not replace it.
Most AI-for-sales coverage is either “AI will replace SDRs” or “use ChatGPT to send 10x more cold emails.” Both miss what good selling actually is: a high-signal series of conversations grounded in research and structured around the buyer’s actual process. The seven prompts below take the parts of selling that compress well with AI (research synthesis, draft generation, deal-stage structure) and protect the parts that do not (the actual conversation, the judgment call on whether to push or back off, the relationship). This is the sales slice of the AI Prompt Library, paired with a connector callout for the CRM and outreach tools sales reps actually live in. For Claude-specific sales prompts see Best Claude Prompts for Sales.
Why do most AI sales-AI workflows produce outreach that gets unsubscribed in two sends?
The default sales-AI loop is to use ChatGPT to draft 200 cold emails a week and call it productivity. The output is plausible, generic, and produces reply rates that round to zero. Worse, it conditions the rep to skip the research step that used to be how they earned the right to be in someone’s inbox.
The prompts below take the opposite approach. Each one requires you to paste in actual research (the prospect’s company, the trigger, the prior conversation, the deal state). The AI scaffolds the writing around your input. Your relationship intelligence stays your moat. If you let AI draft anything customer-facing, run it through How to Edit AI Out of Your Writing first; sophisticated buyers can spot AI rhythm in three sentences and that kills the trust you are trying to build. When a prompt becomes a daily move, graduate it using the Prompt-to-Workflow Ladder.
What are the seven for sales prompts?
Prompt 1
Pre-Call Account Brief
Sales reps walk into too many calls cold. This prompt produces the 90-second brief that puts the rep in the seller’s seat before the meeting.
I have a sales call tomorrow with: COMPANY: [NAME] CONTACT(S): [NAMES AND TITLES] WHAT WE SELL: [OUR PRODUCT IN ONE SENTENCE] WHY THEY ARE TAKING THE CALL: [INBOUND / REFERRAL / OUTBOUND ACCEPTANCE] What I know about them from public sources and prior conversations: [PASTE WHAT YOU HAVE: recent funding, public posts, their tech stack, last email exchange, demo notes, etc.] Produce a 90-second brief with: 1. THE COMPANY IN ONE LINE: what they actually do and how they make money. 2. THE CONTACT'S LIKELY MOTIVATION: why they specifically might care about our product. 3. THE TRIGGER: any recent event (funding, hire, launch, public pain) that makes our product relevant now. 4. THE LIKELY OBJECTION: the single most likely reason this deal does not close, based on what I know. 5. THE OPENING QUESTION: one question to ask in the first 5 minutes that signals I did my homework without being a stalker. 6. THE NEXT-STEP TO PROPOSE: the smallest commitment I should leave the call with. Do not invent facts I did not provide. Flag anything you would research further before the call.
When to use: Night before the call. · Best model: Claude (most disciplined about not fabricating facts).
Prompt 2
Cold Outbound Email Personalizer
Most cold email is templated. The reply rate is 1%. This prompt produces messages that get the better-than-average rep to 5%.
I want to email a cold prospect: PROSPECT: [NAME, TITLE, COMPANY] WHAT THEY DO: [BRIEF] WHAT I KNOW ABOUT THEM SPECIFICALLY: [PUBLIC POST, RECENT NEWS, OBSERVABLE PAIN, MUTUAL CONNECTION] WHAT I SELL: [PRODUCT IN ONE SENTENCE] WHY I THINK THEY SPECIFICALLY FIT: [ONE-SENTENCE THESIS, NOT GENERIC] MY CALL-TO-ACTION: [WHAT I AM ASKING THEM TO DO] Draft a 90-word email with: 1. SUBJECT LINE: 6-9 words, specific to them, not "quick question" or "thoughts?" 2. OPENING SENTENCE: references the specific thing I noticed about them. 3. RELEVANCE BRIDGE: one sentence connecting what they do to what we do, framed as a fit not a pitch. 4. CTA: one specific small ask, not "can we hop on a 30-minute call?" 5. SIGNATURE: clean, no enormous email-block signature. Do NOT use: "reaching out," "touch base," "circle back," "following up," "hope this finds you well," "saw your post and was inspired," "thought you might find this interesting." Do not invent shared connections.
When to use: After 5 minutes of research on the prospect. · Best model: Claude (most disciplined about banned-phrase rules).
Prompt 3
Discovery-Call Question Set
Most discovery calls turn into product demos because the rep ran out of questions. This prompt produces the question stack.
I have a discovery call with: CONTACT: [NAME, TITLE, COMPANY] THEIR STATED REASON FOR THE CALL: [WHAT THEY SAID THEY WANTED] MY HYPOTHESIS ABOUT THEIR ACTUAL PAIN: [YOUR GUESS] WHAT WE SELL: [PRODUCT IN ONE SENTENCE] QUALIFICATION FRAMEWORK MY TEAM USES: [BANT / MEDDIC / MEDDPICC / SPIN / GAP / OTHER] Produce a question stack for the call: 1. CONTEXT QUESTIONS (3): warm up, understand their environment. 2. PAIN QUESTIONS (3): surface what is actually broken in their world. 3. IMPACT QUESTIONS (2): make the pain quantifiable. 4. DECISION-PROCESS QUESTIONS (2): understand who actually decides. 5. TIMING QUESTIONS (1): why now vs next year. For each question: - The question itself in 1 sentence. - The signal I am listening for in their answer. - The follow-up question if their first answer is shallow. Do not lead with the demo. Do not bury qualification questions where they will not happen. Use open-ended questions; avoid yes/no.
When to use: Day before the call. · Best model: Claude is well-suited because of the discipline about not skipping qualification.
Prompt 4
Objection Response Drafter
You just heard “too expensive.” Or “we use a competitor.” Or “we don’t have budget.” This prompt drafts the response that does not sound defensive.
The objection I just heard: [PASTE EXACT WORDS OR PARAPHRASE] The context (what the rep just said, what stage of the deal we are in, what the prospect cares about): [PASTE] What is true about the objection (do not pretend it is not real): [WHAT IS LEGITIMATE] What the objection might be hiding (the underlying concern): [YOUR GUESS] Draft a response: 1. ACKNOWLEDGE: 1 sentence that takes the objection seriously without flattering the prospect. 2. CLARIFY: 1 question that surfaces what is actually behind the objection (because the stated one is rarely the real one). 3. THE REFRAME: 2 sentences that reposition the conversation around what they actually care about. 4. THE BRIDGE: 1 sentence offering a path forward that does not require them to capitulate. 5. THE CHECK: 1 question that lets me read whether the reframe landed. Do not lie about pricing. Do not bad-mouth the competitor. Do not promise discounts I cannot authorize. Tone: confident, curious, never defensive.
When to use: Same call if you can, or immediately after if you need a beat. · Best model: Claude. Tone discipline is the whole game here.
Prompt 5
Multi-Threaded Champion Map
Single-thread deals die. This prompt forces the rep to map the buying committee before pretending the champion is enough.
Here is the deal: COMPANY: [NAME] DEAL SIZE: [APPROXIMATE] MY CHAMPION: [NAME, TITLE] OTHER PEOPLE I HAVE TALKED TO: [LIST WITH TITLES] OTHER PEOPLE I KNOW EXIST BUT HAVE NOT TALKED TO: [LIST WITH TITLES] Deal stage: [DISCOVERY / EVALUATION / NEGOTIATION / VERBAL / CLOSE] Produce a multi-thread map: 1. THE LIKELY BUYING COMMITTEE: roles typically involved in this size of deal in this industry (economic buyer, technical buyer, user buyer, blocker, coach). 2. WHO I HAVE COVERED in each role. 3. WHO IS MISSING. 4. THE BLOCKER: who in this org is most likely to slow or kill the deal, and why. 5. THE COACH: who in this org is most likely to help me from the inside, and what to ask them for. 6. THE NEXT THREE OUTREACH MOVES, in order of priority. 7. THE RISK if I stay single-threaded with my current champion. Do not assume the champion is enough. Most deals die from over-reliance on a single thread.
When to use: Once the deal is past initial discovery; refresh weekly. · Best model: Claude (most disciplined about not letting the rep off easy on single-threading risk).
Prompt 6
Close-Plan Drafter
Most deals slip because the rep does not have a written close plan. This prompt produces one in 10 minutes.
I want to close [DEAL NAME] by [TARGET DATE]. Current state: DEAL STAGE: [STAGE] PRICING DISCUSSED: [YES / NO / IN-PROGRESS] KNOWN OBJECTIONS: [LIST] WHO HAS NOT YET BEEN INVOLVED: [LIST] BUYING PROCESS WE KNOW ABOUT (procurement, security review, legal, exec approval): [WHAT WE HAVE LEARNED] Produce a close plan working backwards from the target date: 1. FINAL WEEK: signature, paperwork, kick-off date. 2. WEEK 2 BEFORE: legal redlines complete, final pricing agreed. 3. WEEK 3 BEFORE: technical evaluation complete, security review submitted. 4. WEEK 4 BEFORE: economic buyer aligned, ROI case made. 5. NOW: the next 3 things I have to do this week. 6. THE RISK FACTOR: the one thing most likely to slip the timeline. 7. THE MUTUAL ACTION PLAN that I should send the prospect to align timelines. Do not draft the MAP as a wishlist. Every milestone needs a clear owner and a clear date.
When to use: As soon as the prospect signals they want to move forward. · Best model: Claude. Discipline about clear ownership matters.
Prompt 7
Lost-Deal Post-Mortem
Reps lose deals and immediately move on. This prompt extracts what to do differently next time, while it is fresh.
I just lost this deal: COMPANY: [NAME] DEAL SIZE: [APPROXIMATE] STAGE WHERE IT DIED: [WHEN IT WENT SOUTH] WHY THE PROSPECT SAID THEY ARE NOT MOVING FORWARD: [THEIR WORDS] WHY I THINK THEY ARE NOT MOVING FORWARD: [YOUR READ] WHAT I LEARNED ABOUT THE COMPETITOR (if they went with one): [PASTE] Produce a post-mortem: 1. THE STATED REASON vs the LIKELY REAL REASON. 2. THE STAGE WHERE THE DEAL ACTUALLY WENT WRONG (it is rarely where it died; usually 2 stages earlier). 3. THE THREE SIGNALS I missed. 4. WHAT I WOULD HAVE DONE DIFFERENTLY if I had the read I have now. 5. THE COMPETITOR PATTERN: if they went with a competitor, what we know about how they evaluate vs us. 6. THE RE-ENGAGEMENT TRIGGER: the event 3-6 months from now that would make this prospect worth re-engaging. 7. ONE TAKEAWAY for my next deal at a similar profile. Do not soften the analysis to protect my ego. Closed-lost is the most valuable data the team has.
When to use: Within 48 hours of the loss. · Best model: Claude or Grok. Both willingly call out rep mistakes.
These work across Claude, ChatGPT, Gemini, and Grok. Claude is most disciplined about not over-claiming and not letting the rep off easy on multi-threading risk. ChatGPT is the broadest for fast iteration on outbound variants. Grok is sharpest for the post-mortem prompt because it will actually call out rep mistakes instead of softening. The thing that makes sales-AI workflows valuable is consistency: running the brief prompt before every call, the personalization prompt before every cold email, the close-plan prompt for every deal in your pipeline.
What is the worst thing you can do with AI for sales?
Three patterns will kill sales-AI workflows fastest.
- Sending AI-drafted cold email without editing into your voice. Buyers know AI rhythm. The well-personalized cold email rate is dropping because AI flattened the format into the same predictable shape. The reps winning right now spend 5 minutes editing every AI draft.
- Letting AI synthesize buyer-committee feedback into a single recommendation. Disagreement among buyers is the most valuable signal you have. AI summaries smooth it away and you lose the read of the deal.
- Outsourcing the post-mortem to AI without your own read. The prompt produces structure; your judgment about why the deal actually died is the thing that compounds across the next 12 months of deals. Spend 10 minutes thinking on your own before reading the AI synthesis.
What if you want to take this further?
Each prompt above takes inputs you paste in. The next move is connecting AI to your CRM and outreach tools so the prompts become near-zero-friction.
Connectors are now standard
Claude, ChatGPT, and Grok all support connectors that let your AI read live data from your work tools (Gmail, Notion, GitHub, Asana, HubSpot, Stripe, and many more) instead of relying on you to paste context. For sales reps this means the AI can read your HubSpot or Salesforce deal data, your Gmail or Outlook outreach threads, your Apollo prospect lists, your Calendly bookings, or your Notion deal-prep wiki.
For sales reps, the connectors worth pairing with these prompts:
- HubSpot connector — reads deal stage, contact properties, prior activity for the pre-call brief and close-plan prompts.
- Apollo connector — reads prospect data for the cold-outbound personalizer with deeper firmographic context.
- Gmail / Outlook connector — pulls prior threads with a prospect for follow-up personalization.
- Calendly connector — for the discovery-call prep, AI references actual upcoming booking and intake-form responses.
- Notion connector — if your team’s deal-prep templates and playbooks live in Notion, AI reads them for consistency.
What are common questions about AI for sales?
Will AI replace sales reps?
AI will replace reps who treat selling as template volume. The work that compresses (account research, draft generation, deal-stage tracking) gets faster. The work that does not compress (the actual call, the buyer’s emotional read, the trust built across a multi-month deal) is more important than ever. Reps who use AI as a co-pilot become more productive. Reps who use AI as a replacement for thinking get out-competed.
Should I use AI for cold email at scale?
No. The high-volume AI-cold-email pattern has tanked reply rates across the industry. The reps winning right now send fewer, better, more personalized emails. Use the Cold Outbound Email Personalizer prompt for each prospect individually, not to spray 200 a day. Volume without personalization is the failure mode you want to avoid.
Which AI tool is best for sales?
Claude Pro ($20/month) is most disciplined about banned-phrase rules and not over-claiming. ChatGPT Plus is broadest. For CRM-integrated AI, your CRM vendor’s built-in AI (HubSpot Breeze, Salesforce Einstein, Apollo’s AI) may cover the routine work; pair with Claude or ChatGPT for the higher-judgment work. Most reps end up with two: one for daily back-office, one for harder analysis.
Is buyer data safe in AI tools?
Paid Claude and ChatGPT plans do not train on inputs and do not retain content beyond the session. Read each provider’s data handling policy. For enterprise deals with NDA constraints, check your client agreement and your company’s AI policy. Never paste contract terms, signed NDAs, or confidential financial data into a chat unless your company has approved that data flow.
Should I tell prospects I use AI?
Most reps do not announce it (in the same way they do not announce which CRM they use). They DO take responsibility for everything they send. If a prospect asks directly: be direct, AI helped structure my research, the read and the message are mine. The direct answer always lands better than the defensive one.
Can AI improve forecast accuracy?
Yes for surfacing patterns in your pipeline (stuck stages, single-threaded deals, missing milestones) that your CRM dashboard does not flag. No for predicting individual deal close dates; AI cannot read whether your champion left the company yesterday or whether procurement decided to delay budget. Use AI for the prompt structure (the close-plan prompt above); the forecast stays your call.
How long does it take to build the sales-AI loop?
Three weeks. Start with the pre-call brief and the cold-outbound personalizer this week. Add discovery question prep and the close-plan drafter in week two. Most reps settle into 4-5 of the seven prompts as part of their daily flow within a month.
The AI Prompt Library · $39
Sales workflows, prompt-paved.
Soon to be 1000+ prompts in Notion organized by use case. The full sales section includes everything above plus prompts for renewal conversations, executive sponsor briefings, pricing negotiation, contract redlines, and territory planning. Plus prompts for every other field. Lifetime access.
Get Smarter About AI Every Morning
Free daily newsletter. Built for people who want to use AI well, not chase every model.
Free forever. Unsubscribe anytime.
Sources to read next?
- Anthropic prompt engineering documentation · official prompt design guide
- MEDDICC framework documentation · the qualification framework referenced in the discovery-call prompt
- Chris Voss: Never Split the Difference · framework for objection-handling and negotiation
- Anthropic: Introducing Connectors · context for the HubSpot, Apollo, Gmail callout
- Force Management: Command of the Message · framework for the close-plan prompt
You might also like
- AI Prompt Library · the full library this post pulls from
- Best Claude Prompts for Sales · the Claude-specific sales playbook
- AI for Sales · the broader playbook
- How to Edit AI Out of Your Writing · the cleanup pass before customer-facing email goes out
- Prompt to Workflow: The AI Ladder · graduate prompts into saved workflows
- Best AI Prompts for Email Writing · for the email-heavy parts of selling
- Best AI Prompts for Meeting Notes · for discovery and demo calls
Want a head start? Book a 2-hour live AI crash course
A private, beginner-friendly session across Claude, ChatGPT, Gemini, Grok, and the wider landscape. Walk away knowing which tools fit your work and how to use them.
Book the 2-hour crash course · $125 →